oaUK Patent Application GB 2320 384 aA (43) Date of A Publication 17.06.1998 (21) Application No 9625632.6 en nrc’ Gots 13788 (22) Date of Filing 10.12.1996 (82) UKCLIEdition P) HaD ORPY 0265 0362 0363 0366 40x (71) Applicantis) ‘GEC-Marconi Limited (56) Documents Cited US 3298656 A (incorporated in the United Kingdom) (58) Field of Search ‘The Grove, Warren Lane, STANMORE, Middlesex, UKCL (Edition ©) HaD DRPY DSPD DSPE DSPU HAT ALY, United Kingdom INTCL® Gots, (72) Inventor(s) Robert Bewes Mobyneux-Berry Peter Alan Langsford Colin Wardell (74) Agent andlor Address for Service G Cockayne GEC Patent Department, Waterhouse Lane, CHELMSFORD, Essex, CM1 20X, United Kingdom (64) Radar velocity measurement (57) _ Radar apparatus stores returns. To determine the approach velocity of a target the target is first classified as a “target of interest" by a thresholding process applied to successive signals to identify potential targets. Doppler processing is then applied only to a selected portion of stored data relating to the target of interest. This invention enables existing moving target indicator MTI radars (1) to be upgraded to give full Doppler capability without the excessive processing capacity required by conventional Doppler radars. r 7 ' 1 ' 1 ' 1 ' 1 i 4 { ' ' t ' t t ' t ' if ' ' = hid ! \ANAGERMENT UNIT} i PRE i 1 (OM), ' 2 ' ' ' 2 1 + 1 1 'Teapan |! ' i} atten |i eens DOPPLER ‘ a 1 ®y VIDEO 9 EXTRACTOR UNIT 1 mc Sore & Ge L ot V v8e0zez aD‘Oo (sao) (naa) > Hols AINA HOLOVHLXS O3dIA u3T1dd0d SIMDAD YALLIWSNVEL yvdvd 6 (awa) JINN LNAWESOVNVIN Wa 1dd0G S1HOd3Y Y31dd00 Se x YASISSHLNAS Se XONSNOSHS (Nig 3ONVY'aINLD) SNOILOSLIG Mv ee ee eee oe \ + ( En) : | eee (NIG SONY'S Motaeted ae 8 uvava \ 101d SNOLLOSISO MVE | av goNVO LW ee J “64 o3aiA2/4 PRI oo4 2 3 4 5 6 7 8 9 10 11 12 PULSE NUMBERRELATIVE SAMPLE PHASE (RADIONS) 4280 426 4240 422 420" 418% +16x +14n 4 t x it &éopn aoa a * STRIKES 30 MILLISECS4/4 ALIODOTAA H3AIddOd 2, 000! 008 009 oor 002 0 00z-_O0b- —S009- _—008- 001, TIAN NHL” + Saundra s “wnaisay — / on st 5 02 a ep 3 ego Z os 249 9 5 se < Ov sp 0s go=m 201A e130uvL = 0FFIA | ziaduvL aah | LOUVL des/wol=dals 0-4 r=-Elv O00l=2@a Sz=F1V o=td ish ‘y Bly10 15 20 1 2320384 LER The present invention relates to apparatus for, and a method of, determining the approach velocity of a target. A basic understanding of the principle of Doppler radars is necessary to the understanding of the present invention. A radar echo from a moving object (or target) is characterised by a frequency shift relative to the radar transmitter frequency. This frequency shift is proportional to the transmitter frequency and also to the rate at which the distance (or range) between the target and the radar was changing at the time when the echo signal ‘was scattered from the target. A positive frequency shift (making the echo frequency higher than the transmitter frequency) occurs when the target approaches the radar and a negative shift occurs when the target recedes. A radar employing the Doppler effect, known as a Doppler radar, needs to resolve and ‘measure small frequency shifts. To do this the radar dwell time in any target direction needs to be sufficient for the degree of resolution required. In theory the best achievable Doppler resolution (in Hz) is the reciprocal of the dwell time in seconds. For centimetric radars concemed with targets on or close to the Earth's surface the Doppler resolution required is usually a few tens of Hz, so dwell times in the range 10 to 50 milliseconiis are commonly used. In the case of radars with mechanically rotated antenna beams the dwell time is limited by (and usually equal to) the time taken by the beam (of finite width) to pass across the10 15 2 target position. This limits the maximum beam rotation rate and determines the minimum time interval between successive looks at a target, on successive rotations, Relatively powerful radars generally employ pulsed transmissions. It is usually impractical to employ single pulses 10 to 50 millisecs long so instead a succession of short pulses are transmitted during the dwell time. Each pulse for which a useful target ‘echo is retumed is called a strike. In order to achieve the required Doppler resolution the ‘echoes from all strikes constituting one look at a target must be in some way coherently integrated when measuring the Doppler shift of the target in question. In practice the (very small) Doppler shifts cannot be measured directly. Instead they are detected by observing small, progressive shifts in phase (with respect to the transmitter signal) on the sequence of strikes obtained during the dwell time. This measurement technique requires that the relationship between the transmitter signal and the local oscillator (and any other frequency reference signals used in the receiver) should be phase coherent and phase stable throughout the dwell time. Even where a radar system is stable and fully coherent, subdivision of the dwell time into a series of very short, impulsive strikes opens the possibility of serious measurement ambiguities. These ambiguities occur because any phase comparison, made on any strike, is inherently ambiguous in respect of multiples of 2 = radians. A measured phase difference of © must therefore be interpreted as an actual difference of (0 + 2Nm.) where Nis an unknown integer, Each plausible value of N implies two different Doppler shifts, ‘one positive and one negative. Usually, very large values of N correspond to implausibly10 15 20 3 high target velocities which need not be considered. However, it is not unusual for there to be 20 or more plausible Doppler shift ambiguities per measurement. Unless these ambiguities can be resolved the Doppler measurement is of little use. One known way to resolve such ambiguities, or at least to reduce their number for each measurement, is to employ transmitter pulses which are unevenly spaced in time during the dwell period. This can be implemented in a variety of ways. One common implementation introduces modest variations in the pulse repetition interval (PRI) on a pulse-by-pulse basis. This is known as PRI (or PRF) stagger (PRF = pulse repetition. frequency). Another known implementation employs a short burst of pulses at one PRF followed by a second burst at a different PRF. Usually, two different PRFs are transmitted alternately in a regular manner. Strike returns from a target must include substantial parts of two or more different bursts in order to ensure a useful degree of ambiguity resolution. Strike retums from one burst are analysed to determine the apparent phase change (@,) per pulse. The interpretation of this in terms of Doppler shift is ambiguous for reasons already explained. Strikes from the second burst are then analysed to determine 8, which will, in general, differ from ©, because the second burst has a different PRI. Doppler shifts calculated from 6, will also be ambiguous, for the same reasons. However, when the set of ambiguous Doppler shifts calculated from one burst are compared with the set calculated from the second burst, one unique value should be ‘common to both sets; this is the correct Doppler shift. In practice, the measurement accuracy of 8, and @, is limited by clutter echoes and noise, by the (relatively small) number of strikes per burst, and by stability imperfections within,10 15 4 the radar system. The sets of Doppler shift values calculated from @, and 6, are therefore subject to small errors with the result that no unique value will match exactly but more than one value may match approximately, leaving the ambiguity problem partly but not fully resolved. However, by careful selection of the radar’s design parameters it is usually possible to ensure that the residual Doppler ambiguities are well spaced in frequency, so that only the correct value lies within the plausible range for targets of interest, ‘Similar principles apply to radars using staggered PRFs, giving generally similar results, although details of implementation are, of course, different. In both cases (PRF staggered and burst mode systems) the Doppler processing is fully integrated into the detection process so that echo signals are resolved and separated into different Doppler bins before detection. This allows echoes allocated to different Doppler bins to be tested separately for detection decisions. When carefully designed, a Doppler radar is very effective and not only positively identifies a moving target from clutter, or stationary objects, but also provides the closing/separation velocity of the target relative to the radar. However it will be appreciated from the above that a great deal of real time processing is required to process all the data received making such radars very costly, especially for radars whose antennas form multiple (eg. stacked) receiving beams, each of which requires a complete Doppler processing channel. One very important advantage of Doppler radars is that they can distinguish moving10 15 20 5 targets, characterised by non-zero Doppler shifts, from signals reflected from stationary or slow-moving clutter, such as tall buildings, terrain features, rain clouds, etc, falling within the radar beam at ranges of interest. Because fast-moving targets of interest are detected in different Doppler bins to those to which echoes from stationary or slow- moving clutter are allocated, the clutter echoes do not compete with the wanted target echoes. This enables a high detection sensitivity to be used for rapidly approaching targets seen in regions of clutter. It also enables the radar to identify and disregard the clutter. ‘The ability to reject clutter echoes is essential to many radar applications but the cost of providing full Doppler processing may be considered too high. Instead, a greatly simplified form of processing, known as moving target indication (MTI) may be used. This rejects echoes from stationary or slow-moving objects, including most types of clutter, but does not otherwise attempt to measure Doppler shifts on target echoes. ‘The essential principle of MTI processing is to subtract the echo signals obtained on successive strikes. Since the PRI is likely to be, at most, a few millisecs, echoes received on successive strikes on stationary clutter or slow-moving targets should be almost identical. Subtracting one from another should leave only a small residue, By contrast, successive strikes on a rapidly moving target will show a relative Doppler phase shift when received. Subtracting these strikes returns should leave a much larger residue. If the relative phase shift approaches m (180 degrees) the “subtracted” signals will in fact add to leave a “residue” of considerably greater amplitude than either signal on its own.10 15 20 6 MTI processing requires the radar transmitter and receiver to be coherent and phase- stable from pulse to pulse. Improved cluter rejection can be achieved by extending the simple subtraction method to cover several pulses (termed double or triple cancellation, etc.); this requires suitable coherence and stability attributes over the duration of several pulses. However, if the approach velocity of a moving target happens to be such that the relative phase shift on successive pulses is exactly 27 (or +2N7, etc) its echo will be cancelled by the subtraction process and any chance of detection will be lost. This is termed Doppler blindness and is closely analogous to Doppler ambiguity in a Doppler radar. ‘The standard cure for Doppler blindness is to vary PRI during the dwell time, by using either PRF stagger of burst mode transmissions, as in a Doppler radar; targets which are blind at one PRI should be seen clearly when a different PRI is used. The method is not perfect but, as with Doppler ambiguity, by careful choice of radar design parameters the lowest residual blind speed can be raised to a velocity which is not plausible for targets of interest. In summary, an MTI radar can achieve clutter rejection performance comparable to that of a Doppler radar having similar detection performance. The MRI radar transmitter must have coherence, stability and PRF stagger attributes similar to those of a Doppler radar but the receiver processing is much simplified and, if desired, can easily be implemented without using advanced digital techniques. MTI technology has been developed to give g00d performance and reliability at modest cost; very many such radars are currently in10 15 20 service. It is an object of the present invention to provide a Doppler radar which requires less processing power than present Doppler radars of similar performance and to provide a way to convert MTI radars to measure Doppler shifts for moving targets. According to the first aspect of the invention there is provided radar apparatus comprising: a transmitter arranged to transmit a series of pulses; a receiver arranged to receive and store successive echo signals from potential targets; a detector arranged to identify echoes of interest by means of a thresholding process applied to data originating from the received signals; and a Doppler extraction processor arranged to process stored data corresponding to an echo of interest identified by the detector in order to determine the approach velocity of a potential target corresponding to that echo. By employing the present invention itis possible to employ the principles used in an MTI radar to identify echoes or targets of interest and then, once such a target has been identified, to process only that data relating to the target in order to obtain the approach velocity. This greatly reduces the quantity of data that has to be handled by the Doppler extraction processor. An “echo of interest” will primarily correspond to a moving target which will be distinguishable from stationary objects and background clutter. However the detector may also take into account other parameters such as strength of received10 15 20 8 signal in determining whether the target is of interest. The stored returned signals extracted from the radar receiver before MTI subtraction has been implemented can be overwritten, therefore only a limited duration of retumed signal need be stored, the stored data is preferably identified chronologically which may be achieved by time tagging the data. In this manner the Doppler extraction processor can have access to the data required to assign an approach velocity to a target of interest, without having to have excessive memory storage. The data can conveniently be stored in RAM in the form of complex video comprising (p) and (q) signals, one in phase and one in quadrature. Preferably the detector comprises means for subtracting successive received echoes and thresholding the difference signal obtained to determine an echo of interest. Thresholding processes can be implemented using relatively little processing power compared to performing Doppler processing every point on an image. ‘Advantageously the Doppler extraction processor is arranged to: retrieve from the stored data the relative phases of successive echoes from a target of interest; compare a range of velocity values with the received phase data; and identify the velocity value that best fits the phase data. (The range of velocity values is preferably limited to a plausible range for a particular target type.) The best fit can be found by calculating a fit error value for each velocity value in the range tested; the set of error values constituting an error function; prominent nulls in the error function are possible values of the target approach10 15 20 9 velocity. The approach velocity corresponding to the deepest null can then be assigned to a target of interest provided the null depth is below a predetermined threshold. To compare null depths more accurately the Doppler extraction processor may determine the best fit error values by using a finer range of velocity steps in the region of prominent nulls in the error function. ‘The above process permits the error function to be generated relatively quickly by using a fairly coarse velocity step, thresholding it to find prominent nulls (of which there will typically be several) and then to recalculate small sections of the function in the vicinity of these nulls, using a relatively fine step in order to determine which of the coarse nulls is in fact the deepest and whether it crosses the final acceptance threshold. In an alternative to the above, the Doppler extraction processor can be arranged to apply a weighted “least sum of squares fit” to a null to determine the minimum of the null. Performing a weighted “least sum of squares” fit (using known algorithms) enables the absolute minimum null value to be determined for each candidate null region. One advantage of this is that the “best fit” algorithm is not based on any assumption that the reference phase 6 (1) of the central strike has been measured with negligible error. This allows deeper and more reliable null values to be found in cases where the signal to noise ratio (SNR) is relatively poor. It is preferable that the pulse repetition interval is varied, either on a pulse by pulse basis or on a burst of pulses basis, since this reduces ambiguity in the Doppler values.10 15 20 10 ‘The output of the Doppler extraction processor may advantageously be used to verify the target as being a “target of interest” if a velocity can be determined. This need not necessarily require the velocity value itself to be output but could simply determine that the relative phase values of a number of successive echoes are such that they correspond to a plausible velocity value. Exploitation of Doppler data by target tracking algorithms is well known, especially where the tracks of independent targets happen to cross. However Doppler measurements ‘obtained according to the Invention can also be exploited by false target filter algorithms. ‘The Invention provides a good, unambiguous Doppler measurement, characterised by a unique, deep null in the error function, only in cases where a sequence of target strike echoes is characterised by a clear, orderly phase progression. It is less probable that clear-cut Doppler measurements will be obtained from (for example) a sequence of inregular, distributed strikes due either to storm-whipped sea clutter or to large flocks of birds. Failure to extract a clear Doppler measurement for a reasonably strong detection ‘not suppressed by MTI thus indicates that the target is probably due to distributed moving clutter, ie. either false or unwanted. Most known methods of identifying unwanted false target detections rely on comparing detections obtained over two or three looks on successive antenna rotations. By comparison, in addition to providing Doppler measurements for wanted targets, employing the Invention can provide an “instantaneous” test for identifying certain important (and troublesome) categories of false target.10 15 20 u Preferably the apparatus further comprises means for identifying slow moving clutter at Teast in part in dependance on the approach velocity determined by the Doppler extraction Processor, means for estimating the amplitude of the signal returned by said clutter, ‘means for generating a synthetic clutter signal and subtracting this from the received data corresponding to said clutter to obtain a residual echo signal, wherein the Doppler extraction processor determines the approach velocity of a target corresponding to the residual echo signal. The synthetic clutter signal is advantageously subtracted from the received signal at a plurality of different phase shifts in order to determine the relative phase which provides a minimum residual signal, whereby the Doppler extraction processor determines the approach velocity of a potential target corresponding to the minimum residual signal. This enables echoes from a slow moving source such as would be generated by chaff to be identified and the return signal from the slow moving source to be substantially deleted in order to expose the “hidden” target, for example an aircraft, which may already have reported by an MTI thresholding process. According to a second aspect of the invention there is provided a method for obtaining approach velocity information from a moving target indicator radar (MTI) radar, the method comprising storing successive retumed radar signals (abstracted before MTT subtraction), retrieving from the store data relating to a target of interest identified by the MTI radar, and processing that data to determine the approach velocity of that target. ‘This enables approach velocity information to be extracted from data already present in existing MTI radars, without requiring excessive processing power required by conventional Doppler radars.10 15 20 12 According to a third aspect of the present invention there is provided a method of determining the approach velocity of a target, the method comprising: transmitting a sequence of pulses of electro magnetic radiation; receiving and storing data from successive echoes from potential targets; identifying echoes of interest by applying of a thresholding process to data originating from successive received echoes from a potential target; and for an echo of interest determining the approach velocity of the corresponding target. The invention will now be described by way of example only with reference to the accompanying figures of which: Figure 1 is a functional block diagram of the components of a radar system in accordance with the present invention; Figure 2 illustrates an exemplary pulse repetition interval; Figure 3 illustrates the phase ambiguity associated with a returned series of echoes; and Figure 4 illustrates a typical phase ambiguity error function for a retumed series of echo. The invention is described below as it would be applied to a mediunvlong range pulsed MITT radar with a rotating antenna, transmitting in the centimetric band and using staggered PRF in order to overcome Doppler blindness. This type of radar is typical of those in which the invention is of greatest value. The description explains first the10 15 13 principle of the invention before discussing how it is to be implemented. Application of the invention to the receiver of a conceptual existing MTI radar is illustrated by the outline Block Diagram shown in Figure 1. The existing radar transmitter would remain unaltered. Figure 1 distinguishes between equipment functions 1 , comprising the existing MTI radar, and equipment functions 2 added by the invention. The Doppler processing hardware to be added may consist of a digital processor supported by digital memory media. Random access memory (RAM) is suitable for temporary storage of the radar signals being processed; non-volatile memory, eg. using magnetic media, is required for long-term storage of processing and system management algorithms. In Figure 1 the existing MTI radar is controlled by a frequency synthesiser and synchronising unit 3. Because the MTI function requires the radar transmitter 4 and receiver 5 to be coherent and phase stable for the duration of several pulses an MTI radar will include a synchronisation unit 3. The Doppler measurement system makes use of PRF and other synchronisation signals generated by the unit 3. ‘The receiver 5. supplies echo signals in video form to the MTI canceller and detector unit 6 . Efficient clutter cancellation, avoiding both Doppler blindness and phase blindness normally requires double or triple cancellation performed on complex video signals. ‘Complex video comprises two signals, one (p) in phase and the other (q) in quadrature, In some MTI radars the two parts of the complex video signal may be multiplexed in10 15 20 14 some manner. However, for simplicity of explanation it is assumed here that the two parts of the video signal are presented in separate channels. Many MTI radars transmit chirped or phase-coded waveforms and implement pulse ‘compression within the receiver. Pulse compression is normally implemented before the complex video waveform is presented to the MTI canceller and detection unit 6 Provided this is the case it makes no difference (for Doppler measurement) whether or not pulse compression is used. MTI subtraction processing may be performed by either analogue or digital means. Digital implementation is normally preferred where multiple subtraction is to be performed on complex video signals. The invention will therefore be described in relation to complex video signals presented in digitised format. A digital sampling rate of, say, 10 MHz would be typical for an existing MTI radar of the class considered. ‘Where only analogue video is presented, one method of applying the invention would be to digitise the video signals at the input of the added Doppler extraction processor. In such cases it is desirable to use a digital sampling rate which is phase-locked to the master oscillator governing the synchronisation unit 3. ‘The MTI canceller and detector unit 6 subtracts echo signals due to stationary clutter and slow-moving targets before thresholding the residual video to obtain raw detections of targets of interest. In order to do this it must store and subtract video echoes received over several pulse intervals. This means that there is a short time lapse between reception of the first strike on a target and the time when unit 6 is able to report a raw detection